Estimation of the Urban Temperature Reduction Effect in Low-rise Residential Areas using CFD Simulation based on the Shape of the Green Space

1.1. Research background and purpose

Currently, the rapid urbanization being experienced in cities is causing various problems because the green areas that were secured are insufficient. Among them, the urban heat island effect has received increased attention. This effect refers to a phenomenon wherein the center of a city exhibits higher temperatures than the suburban areas. This is caused by a reduction in the green space owing to changes in the ground cover, an increase in the reflectance due to a higher ground surface pavement and larger total floor area of the buildings, carbon dioxide emissions from human industrial activities, and the generation of artificial heat in cities[1] . This urban heat island phenomenon affects the overall human life including increased energy consumption in cities, diminished quality of life for vulnerable societal classes, an increase in the number of patients with heat-related illness, and intensified air pollution[2]. Particularly, this phenomenon is more significant in old downtown areas where the population density increases over a short period of time without proper planning on the use of the existing urban infrastructure.

Land use in cities is known to affect both urban climate control and the mitigation or intensification of urban heat islands. According to a related study, the heat island phenomenon is significant in densely-populated areas owing to high-density development, and an abundance of green space is effective in mitigating the heat island phenomenon[2]. In addition, insufficient quantities of green spaces in a city results in a lack of rest areas for citizens, a reduction in the habitats used by organisms, and the exacerbation of the residential environment in the city. Generally, urban green spaces affect urban climate control; however, studies on the effect of each element, including the size, shape, and location of urban green spaces, on climate control are insufficient. In this situation, minimum quantitative criteria such as the criterion for the area of the urban green space per capita have been suggested, but specific and qualitative criteria such as the shape, location, size, and facility criteria, have not been presented, making it difficult to perform planning in the field.

As such, the aim of this study is to empirically study the urban temperature reduction effect based on the shape of the urban green space, particularly in low-rise densely-populated residential areas in old downtown areas. This is because the urban green spaces in such areas have different shapes and sizes since they are relatively insufficient and developed later. Therefore, these research results can be utilized as important references when developing additional urban green spaces in old downtown areas.

1.2. Research method and scope

The spatial scope of this study was set to 1034-8 Joyang-dong, Sokcho-si, Gangwon-do; 16 Hapjeong-dong, Pyungtaek-si, Gyeonggi-do; and 11-10 Bisan-dong, Seo-gu, Daegu, South Korea. These areas were chosen owing to the following reasons. First, objective meteorological measurements, which are essential for this study, can be obtained from the automatic weather system (AWS) of the Korea Meteorological Administration (KMA), which is installed nearby. Second, these areas are low-rise densely-populated residential areas that were developed long ago. Third, these areas contain more than five types of urban green spaces that were later naturally or artificially developed within the scope of the simulation (1km²).

The temporal scope of this study, which is the period within which the latest verified meteorological data are available, was June to August of 2021, a midsummer period with the highest average temperature in the cities. The basic urban- and building-type data such as the topography, shape and height of buildings, and road types, were based on the basic spatial information (2015) provided by the National Geographic Information Institute (NGII), which was updated based on the latest aerial image data.

In this study, a theoretical review of the literature related to the influence of urban green spaces on the urban heat island effect was conducted. Subsequently, modeling and meteorological data acquisition were then performed in the target areas, and based on the results, the proposed research model was verified using a computational fluid dynamics (CFD) simulation. A CFD simulation was then conducted, the temperature correlations were analyzed using the area, perimeter, and shape index of each green space based on the temperature data extracted from the CFD simulation results, and the influence range of the temperature reduction effect was analyzed for each green space. Finally, the implications of developing urban green spaces in the future were derived based on the results.

2.1. Review of the theory

2.2. Review of previous studies

The previous studies on the changes in the microclimate due to green spaces can be mainly classified into three categories. The first involves studies that employed the direct measurement of microclimate changes using AWS, the second involves studies wherein the change in the surface temperature was measured using aerial and satellite images based on methods such as Landsat and light detection and ranging (LiDAR), and the third involves studies wherein research was conducted using a CFD simulation software and based on an assumed green space type and shape. Table 1. summarizes the review of the previous studies based on these three types.

Table 1.

Summary of the literature review

First, J. K. Kim et al. (2020) analyzed the factors that affect the temperature of the Minjumaru Square, a green space in University C, using AWS measurements. This is an important study because the temperature changes were measured in a small area according to the cover, space, and vegetation[7]. G. W. Lee and Y. N. Jeong (2018) developed a microclimate map using KMA AWS by directly measuring the AWS values and analyzing the microclimate situation in declining areas. They revealed that a temperature change of more than 2℃ can occur within a radius of 500 m depending on the space shape and cover condition[8].

However, studies that use AWS measurements are faced with the difficulty of solving the error caused by the accuracy of measurements and particularity of the measurement time, and only measurements in a small area can be obtained. Moreover, they can only measure the current climatic environment.

Piao, Yulin and Y. H. Kim (2021) analyzed the temperature reduction effect and its range of influence based on the area, shape, and perimeter of urban green spaces in Cheonju using Landsat8 satellite images and the ArcGIS software. They classified the temperature reduction effect according to the area, shape, and perimeter of the green spaces, and identified the most efficient green area through regression analysis considering the temperature reduction effect[9]. K. J. Kim et al. (2018) analyzed the temperature reduction effect in Seoul according to the green space type and subsequent building energy reduction effect using Landsat8 satellite images, making it an important reference for this study[10]. Typically, the use of satellite or aerial images for analysis enables the analysis of larger spaces compared to using AWS. However, it is difficult to simulate the atmospheric temperature even though it is possible to measure the land surface temperature.

For studies that used CFD, W. K. Bae and J. H. Kang (2022) analyzed the temperature change according to the arrangement of the pilotis and green spaces in an apartment complex in Seocho-gu using the ENVI-met software. This is an important study is because the influence of rooftop green spaces, wall green spaces, and pilotis on the microclimate was examined by applying various design scenarios[11]. J. H. Kim et al. (2012) analyzed the changes in the thermal environment of an apartment complex in Suwon before and after the construction of new buildings according to the change in the green area ratio using Star-CCM+, making it an important study for reference significant[6]. S. H. Hong et al. (2020) analyzed the effects of the building height and roadside green spaces on the PM concentration using an ENVI-met simulation[12]. S. H. Kim and W. K. Bae (2021) conducted research on the fine dust reduction effect based on a green space in the alley of a low-rise residential area using ENVI-met. They examined the effect of the green space arrangement on the improvement of the low-rise residential area environment[5]. G. W. Lee (2021) analyzed the effect of developing small-scale green spaces on the microclimate of an urban regeneration area in Daegu using Star-CCM+[3].

There are also a number of overseas studies that have analyzed the temperature reduction effect of green spaces using CFD simulations. Among them, Yupeng Wang (2015) compared the temperatures estimated using ENVI-met with the actual temperature of the downtown and airport areas of Toronto, Canada, from January 16^th to 17^th and from July 17^th to 18^th, 2013. They verified the temperature reduction effect of the cool pavement, cool roof, green areas, and UHI for the detached house, middle-rise, and high-rise areas in the city[13]. Wanlin Liu et al. (2022) estimated the microclimate of an area that had water bodies and green spaces in Dujiangyan, Chengdu, the capital of Sichuan Province in China. They employed CFD simulations using Fluent to acquire meteorological data, maps, and remote sensing images in January and August 2018, and calculated the greenhouse gas emissions from residential areas by inputting the estimation results into HDY-SMAD, which estimates the building energy consumption. They found that the appropriate area ratio between the distributed green spaces and water bodies is 3:1, and that 1:2 is appropriate for the summer[14].

In all previous CFD-related studies, the expected effect was estimated using scenarios based on conditions expected in the future. Therefore, CFD is more beneficial than the previous two research categories because the expected effect can be measured in advance under specific assumptions.

2.3. Research issue and setting the hypothesis

As aforementioned, there are several studies on temperature reduction using green spaces; however, very few have employed the conversion of the area and shape of each green space into quantitative indices and analyzed the temperature reduction effect accordingly. Based on these studies, the following research issues and hypotheses were derived:

1. The size of the green space affects the temperature reduction effect.

2. The temperature reduction effect of the green space is different depending on the shape of the green space.

3. The influence range of the temperature reduction is different depending on the size, perimeter, and shape of the green space.

3.1. Selection of the target areas

To verify the abovementioned research hypotheses, three target areas were selected based on the following criteria. First, the area must support easy objective data acquisition due to the short distance from the KMA AWS. Second, the area must comprise old low-rise densely-populated residential areas. Third, various types of natural or artificial green spaces must be present within the simulation target range of 1km². Based on these criteria, the following areas were selected: 1034-8 Joyang-dong, Sokcho-si, Gangwon-do; 16 Hapjeong-dong, Pyungtaek-si, Gyeonggi-do; and 11-10 Bisan-dong, Seo-gu, Daegu, South Korea.

The first target area, 1034-8 Joyang-dong, Sokcho-si, Gangwon-do, is close to the East Sea and Lake Cheongcho (Fig. 1.). This area is dominated by type-2 general residential areas mixed with some general commercial areas located around the Donghae-ro six-lane road. This area represents the residential areas created long ago around the port.

Fig. 1.

Site 1: Sokcho (Source: https://map.naver.com/(2022.12))

The second target area, 16 Hapjeong-dong, Pyungtaek-si, Gyeonggi-do, is close to Pyungtaek City Hall and dominated by type-2 general residential areas (Fig. 2.). It comprises a low-rise densely-populated residential area that retains the old urban structure even though several large facilities such as Sosabeol Leports Town have been constructed.

Fig. 2.

Site 2: Pyungtaek (Source: https://map.naver.com/(2022.12))

The third target area is 11-10 Bisan-dong, Seo-gu, Daegu, which includes the ‘Indongchon urban regeneration’ area. This is an area that was developed based on the third industrial, dyeing industrial, and Seodaegu industrial complexes during the boom period of the textile industry in Daegu (Fig. 3.). It was a residential area around industrial complexes where a large population once lived; however, population outflow has been recently accelerated by changes in the industrial structure[3].

Fig. 3.

Site 3: Daegu (Source: https://map.naver.com/(2022.12))

3.2. Setting the target green space

These areas highlight the declining trend in the characteristics of residential areas in Korea, where grid systems and unorganized irregular urban space types are mixed. Moreover, since the green spaces were created using the existing urban structure, various types of special green spaces such as linear green spaces are present. Therefore, the effect of developing parks and green spaces and controlling various variables in the target areas to determine the implications of including green spaces in domestic urban regeneration projects or urban restructuring processes was verified in this study.

3.3. Shape index of the green space

The shape index was derived by applying the abovementioned shape index calculation formula to the green blocks employed in this study to quantitatively analyze their spatial characteristics. This index was introduced by J. H. Suh et al (1999) for landscape analysis[15, 16], and was applied by Piao Yulin and Y. H. Kim (2021) to analyze the temperature distribution based on the shape of the green space using Landsat8 satellite images of Cheongju[9]. This index is defined using Eq. 1.

Table 2.

Modeling results for each green areas

D: shape index
S: green block perimeter
A: green block area

The shape index, which indicates the shape ratio, is 1 for a circular target shape and approximately 1.13 for a square one. As it increases, the shape is longer similar to a rod or closer to irregular shapes such as the star and starfish shapes. Therefore, the shape index can be used to comprehensively express the morphological characteristics of components[9]. Additionally, the area and perimeter were extracted based on Rhino modeling. Table 3. shows the extracted area, perimeter, and shape index calculated using these values.

Table 3.

Index for each shape of the green area

3.4. Analysis method

2) Modeling method

A modeling area of 1 x 1 x 0.05 km was used. For the domain, the width and length were increased by five times and the height by ten times based on research criteria determined by Hu and Wang (2005)[17]. In addition, a trimmed mesh, one of the basic grid models, and turbulence were generated when developing the model to express the fluid flow. CFD simulators use various techniques to apply the grid, which is the basic unit of flow analysis, to the shape of the model being simulated. Among them, trimmed mesh applies the grid to fit the shape of the model by reducing its size based on a rectangular grid. For areas that require a more specific simulation and observation, the grid was generated by applying three prism layers to obtain more specific research results. The detailed modeling process is summarized Fig. 4. Consequently, approximately 11.8 million grid cells were generated for the target area in Sokcho, 6.1 million grid cells for the target area in Pyungtaek, and 31.4 million grid cells for the target area in Daegu.

Fig. 4.

Modeling process

3) CFD simulation model

The temperature reduction effect of the green spaces was measured using a CFD simulation model. For the model, Navier-Stokes equations were applied as governing equations, and a realizable K-epsilon turbulence model, which is a representative turbulence model, was applied.

Table 4. shows the boundary conditions entered into the simulation for this analysis. Meteorological conditions were assumed for each target area based on the average wind speed, average wind direction, and average temperature measured from July 1 to August 31 in 2021, as shown in Table 4.

Table 4.

Boundary Conditions

The thermal radiation characteristics of the surfaces, which were defined using the land cover conditions, are shown in Table 5.[18]. Moreover, the thermal radiation characteristics were expressed using the density (ρ), specific heat (Cp), thermal conductivity (k), and emissivity (ε), and the thickness of the ground surface was set to a sufficient depth (4 m), which is not affected by the temperature change of the outer surface. For the green space, the latent heat of vaporization was considered.

Table 5.

Properties of the surface material

4.1. Simulation results for each target area

A model that was previously-verified by G. W. Lee (2021) was used to simulate the temperature reduction effect of each green space and its influence range[3]. Fig. 5. shows the simulation results of this model based on the meteorological data for each target area. For an easy comparison, a temperature range of 25.5 to 40℃ was applied.

Fig. 5.

Simulation results for Sokcho

Fig. 6.

Simulation results for Pyungtaek

The results show that Sokcho is mainly dominated by southeasterly winds, and the spread of cold air occurs around the main green spaces and lakes.

In Pyungtaek, easterly winds are dominant, and the temperature of the hot air coming from the east decreases after coming into contact with the green spaces between the urban structures inside the target area. However, in the densely-populated central and western areas, the wind speed inevitably decreases and the temperature increases owing to artificial heat.

Finally, the highest overall temperature is observed in Daegu, where a low wind speed and high temperature are also observed owing to the high density of residential areas in the target areas. Moreover, southwesterly winds are dominant, and the wind flows along the linear green space near the Gyeongbu Line railroad to provide the temperature-reduction effect. However, a high temperature is observed in this region because of the difficulty in introducing wind into the internal structure owing to the high residential density.

Fig. 7.

Simulation results for Daegu

4.3. Green blocks temperature reduction results

1) Analysis of the temperature reduction effect of the green blocks

Based on the method described above and models developed according to the types of the selected green spaces, simulations were performed and the temperature change measurements were analyzed. For the temperature of the green space, a buffer was included in the perimeter of each green block and the average temperature in each buffer range was calculated. Because the temperature measurements are not accurate if buildings are present in the buffer range, the temperatures of the areas with buildings were excluded. Fig. 8. shows a graphical representation of the temperatures of each green space based on the buffer distance.

Fig. 8.

Temperature reduction effect range for the green area block

The temperature reduction effect of the green spaces was first analyzed using graphs. In all graphs, the temperature was lowest at the center of the green space and increased as the distance from the periphery of the green space increased, indicating a reduction in the temperature reduction effect, The reduction effect was shown to range from a minimum of 1–2°C to a maximum of 4–5°C. It was also observed that the temperature reduction effect has an influence at a distance of 10–300 m from the periphery of the green space.

Additionally, the results for G10 and G18 show that the absolute temperature of the green space decreases as the green space area increases, and that the temperature reduction effect in the surrounding areas is larger as the green space area increases.

The results for G3 to G7 show that the absolute temperature tends to decrease as the slenderness of the shape of the green space increases. However, for the cases where the absolute green space area is large, the absolute temperature is not exceeded.

When the green spaces excluding G3 to G7, G10, and G18 were compared, it was observed that the temperature reduction effect in the surrounding areas decreases as the slenderness of the shape of the green space increases. This effect is larger when the shape of the green space is close to a square shape.

2) Analysis of the temperature reduction effect according to the morphological characteristics of the green space

Table 6. shows the analysis of the correlations between the temperatures for each buffer distance and variables for each spatial feature.

Table 6.

Correlation coefficients and R2 between the shape index and temperature

The area and perimeter show negative correlations, indicating that the temperature decreases as the area and perimeter increase. Moreover, the shape index exhibits negative correlations, confirming that the temperature decreases as the shape index increases, that is, as the shape becomes more complex.

For the R² of each spatial feature, the R² value for the area tends to increase up to a distance of 100 m before decreasing. This means that significant levels are influenced by the green space area for a distance of up to 100 m.

For the perimeter of the green space, the explanatory power was highest at the center and decreased until a distance of 25 m before increasing until 300 m. This means that the temperature of the green space and its vicinity can be determined based on its perimeter, and that the explanatory power decreases as the distance increases.

In the case of the shape index, R² is relatively small compared to the area or perimeter of the green space. However, the phenomenon wherein the temperature at the center decreases as the shape of the green space becomes simpler can be explained.

When these three indices were analyzed, it was found that the green space area is an index that can be used to determine the influence range of the temperature reduction effect of the green space. Particularly, the perimeter of the green space is considered an important variable used to determine the temperature reduction effect of the green space at the center and its vicinity. Moreover, when the slenderness or complexity of the shape increases, the temperature reduction effect is observed; however, its length of influence is short and is only effective in the vicinity. This is regarded to be due to the reduction in the effective area by half in the calculation formula of the shape index, which influences the temperature reduction effect caused by the green space area over a long distance.

Therefore, the green space area is an important factor when analyzing the temperature reduction effect of the green space over a longer distance. Additionally, it is more effective use a slender or irregular green space over a square one when maximizing the perimeter of the green space to maximize the temperature reduction effect over a short distance. This result confirms the findings of G. W. Lee (2021) that the temperature reduction effect of linear green spaces is larger over a short distance[3].